Vacancy PhD researcher: Artificial agents in dynamically changing environments

(08-10-2019)

The research will be performed with the groups of Prof. Yves Van de Peer (VIB) and co-supervised by Prof. Pieter Simoens (UGent IDLab), dr. Yara Khaluf (UGent IDLab) and Prof. Pieter Audenaert (UGent IDLab).  This work builds on the expertise of these groups in systems biology, collective intelligence and networking. Your work will consist of applying bio-inspired evolutionary techniques into behavioral controllers of artificial agents, performing simulations and analyzing topologic effects in the Gene Regulatory Network. More details are provided below.

Research Context  

One of the important challenges in the field of artificial intelligence is the development of systems that can adapt to a changing environment. Under a dynamically changing environment, a solution that is optimal at a certain time might be different from an optimal solution at a later time. 

Being naturally occurring examples of adaptive systems, biological systems provide an important source of inspiration for designing artificial systems. The molecular mechanisms underlying the adaptability of biological systems are Gene Regulatory Networks (GRNs), which are composed of interacting genetic entities such as genes and proteins. Several bio-inspired systems have been developed that use an artificial genome (AG) and a corresponding controller, usually a network structure represented by an Artificial Neural Network (ANN). For instance, a ‘gene’ will define a node in the ANN, but this node will find and connect with other nodes in the ANN based on the given conditions. Other types of controllers are however also possible. Recent approaches use biologically realistic versions of an AG that mimics features of real biological genomes.

In this PhD, we will investigate evolutionary strategies that allow for short-term adaptation in drastically changing environments (e.g. a sudden drop of food). To survive such abrupt changes, the controller must adapt in only a few generations to novel behavior strategies that often deviate significantly from the controller before the occurrence of the environmental change. The starting point is a duplication of the GRN. In nature, it has been observed that mutation through whole-genome duplication (polyploidy) is usually a dead-end, unless the environment has drastically changed (e.g. vulcano eruption). 

The goal of this PhD is to explore the particular network structures that appear in GRN after whole-genome duplication and to better understand the network dynamics that occur in the subsequent mutations (and the resulting controllers). Although the focus is on mathematical modelling, network analysis, simulation and implementation of controllers for artificial agents, you will work in close collaboration with biologists to foster mutual exchange of ideas. The use of artificial agents to study biological phenomena is a complementary approach to biological studies and theoretical models. In comparison to biological studies, artificial agents have the advantage that the evolution of hundreds of generations of controllers can be completed within hours or days.

Offer

A full-time funded PhD scholarship. You will aim for a PhD degree in computer science engineering.

Requirements

  • You have a degree as Master of Science/Engineering, preferably in Computer Science, Informatics, Mathematics or Statistical Physics. Students close to their graduation are also welcome to apply. You have a solid academic track record (graduation cum laude or grades in the top 15% percentile).
  • You have a strong interest in bio-inspired algorithms, complex adaptive systems and network topological patterns (topologies such as scale-free networks, triadic motifs). Proven experience in one or more of these topics, e.g. via master thesis or projects, is a plus (but not a necessity).
  • You are willing to closely collaborate with biologists to better understand the governing principles, and to report your results and insights to them in order to foster scientific progress in both domains.
  • You speak and write English fluently.
  • You have good communication skills and you are a team player.

Interested?

Apply with motivation letter, scientific resume, abstract of your master thesis, diplomas and detailed academic results (courses and grades), relevant publications, and at least one reference contact. This information, as well as possible questions, must be sent to Prof. Pieter Simoens at pieter.simoens@ugent.be 

After the first screening, suitable candidates will be invited for an interview (also possible via Skype) and may get a small assignment. Applications will be screened as soon as they are received. The position remains open until the vacancy is filled.

Estimated starting date: January-February 2020.

 

Read more articles about: